EP1397699B1

EP1397699B1 - Tapped delay line high speed register

Info

Publication number: EP1397699B1
Application number: EP01987017.9A
Authority: EP
Inventors: John J. Drinkard; Ed Oltman; Galen Knode
Original assignee: Scientific Technologies Inc
Current assignee: Omron Scientific Technologies Inc
Priority date: 2000-11-28
Filing date: 2001-11-20
Publication date: 2013-10-09
Anticipated expiration: 2021-11-20
Also published as: WO2002045213A3; WO2002045213A2; EP1397699A2; US6493653B1; ES2441205T3; AU2002239275A1

Description

FIELD OF THE INVENTION

The present invention relates generally to timing systems to measure short time intervals, and more particularly to timing systems suitable for time of flight pulse measurements such as found in systems used to guard protected equipment.

BACKGROUND OF THE INVENTION

In many applications it is necessary to know the distance between two points. Although knowledge of distance per se can be used to make a range finder, in other applications knowledge of distance can be used to protect a zone against intrusion. A factory may have robotic or potentially hazardous equipment that is to be protected from outsiders. A system that can measure the distance between such equipment and a perimeter region around the equipment can sound an alarm or turn-off the equipment if anyone approaches closer than the periphery of the protected zone. In this fashion, outsiders are protected against harm from the equipment, and any operators using the equipment are protected from harm by being startled or otherwise disturbed by outsiders.
Fig. 1 depicts a generic so-called time-of-flight system 10 used to calculate the distance X between system 10 and a target (TARGET). System 10 may be located adjacent robotic or perhaps hazardous machinery in a factory where an alarm is to be sounded or the machinery turned-off if anyone approaches closer than distance X.
Typically system 10 includes a trigger generator 20 that creates a pulse train that is input to a transmitter (XMTR) 30, such as a high speed laser, that broadcasts a pulse via a suitable lens 40. The broadcast pulse 50 radiates outward at the speed of light, and at least a portion of the radiation may contact the surface of the target, and be reflected back toward system 10. The reflected-back radiation 55, which also travels at the speed of light, is detected by an appropriate transducer 60 (e.g., an optical lens) and photodetector 70. In a zone protection application, a mirror within system 10 mechanically rotates in a plane such that transmitted pulses scan the protected region, and return pulses are detected from this region. The protected region may be defined as a swept arc centered on the equipment to be protected, and extending outward with a radius of at least X. Typically the laser transmitter is triggered or pulsed with a known frequency in synchronism with mirror rotation such that detected return pulses can be correlated with an angle of emission, to locate the angular position and range of the intruding object. In such applications, any target (TARGET) within range X within the swept protection zone is presumed to be an intruder. Note that X may be a function of scan angle in that the guarded perimeter need not be defined by a swept arc.
As indicated in Fig. 1, there will be a phase or time shift between corresponding portions of the radiating pulse energy 50 and the return or reflected back radiation 55. Thus, at time t0 a first radiated pulse transitions 0-to-1, but the same pulse upon detection (denoted now P1') will have its leading edge transition 0-to-1 at time t1+Tw later than t0. A high speed counter logic unit 80 within system 10 then attempts to calculate the difference in time between t1+Tw and t0. Tw is a signal strength dependent term that is sometimes called "timing walk".
Within unit 10, detected return pulse P' is amplified and coupled to a comparator to determine the return pulse transition timing. Return pulse transition timing is typically dependent on the strength of the return pulse, which in turn is determined by object reflectivity and range. In Fig. 1, T1 is the delay corresponding to the physical separation between system 10 and the object or target, whereas Tw is the timing walk strength dependent term.
Typically unit 80 includes a high speed master clock 85 (CLK) and a high speed counter 90 (COUNT). At time t0, as determined by a START pulse associated with the beginning of an output emission 50, counter 90 begins to count clock pulses. At time t1+tw, when pulse P1' is detected, counter 90 is halted upon receipt of a STOP pulse, and the count value is determined.
Typically Tw is strongly dependent upon the signal response of the transmitter and receiver circuitry and must be characterized. Correction values are determined over a range of P1' signal strengths and are stored in a table.
The values stored in the correction table are indexed by detected signal strength and may be used by a system control circuit to extract the value t1. Thus, prior art systems that employ time-interval counters typically will use a peak-detector or signal integrator.
Once t1 is known, a measure of distance x given Δt = (t1 - t0) is determined by the following equation: $x = \frac{c • Δ t}{2}$
where c = velocity of light (300,000 km/sec).
Within system 10, generating, transmitting, and receiving pulses can be straightforward. But it can be challenging for system 10 to resolve the distance X within a desired measurement granularity or tolerance. For example, to measure distance with a resolution granularity of about ± 5 cm requires a 3 GHz counter. Such high speed devices are expensive and typically consume several watts of electrical power.
An alternative approach would be to replace the function of high speed clock 85 and high speed counter unit 90 with a high speed analog-to-digital converter. However high speed analog-to-digital converters are relatively expensive.
Yet another approach would be to replace units 85 and 90 with a transient recorder, perhaps inexpensively implemented using common CMOS fabrication processes. Transient recording could be extremely fast yet would not consume excessive electrical power. One prior art transient recorder technique is described in a Univ. of Calif. At Berkeley 1992 M. Sci. thesis entitled "A Multi-Gigahertz Analog Transient Recorder Integrated Circuit" by S. A. Kleinfelder. Kleinfelder's thesis described a tapped, active delay line using an array of storage capacitors. The capacitors stored samples of the detected return pulse P1' at specific delay times that were set by the delay of each element in the delay line.
Kleinfelder's approach appears ideal in that it presents a fully digitized representation of the delayed pulse (or multiple pulses), at relatively minimal cost. Further, no thresholding of the analog return pulse is necessary, and range distance may be computed using an algorithm that takes into account the full pulse shape. The latter is important in determining target range, independently of the strength of the return pulse P1.
Unfortunately, in practice Kleinfelder's system is difficult to implement because of the large amount of data that must be processed in a relatively short time. Further, it is necessary to characterize performance of the active delay line and particularly the storage capacitors and analog-to-digital converter circuitry over process, temperature, and voltage variations.
US 6,137,566 A discloses an apparatus for receiving signals from a photodetector in a photodetector array. The apparatus has a number of comparators connected to the photodetector. Each photodetector in the array has a similar circuit connected. Each comparator compares the output of the photodetector with an adjustable threshold level. In this document, data acquisition can be started upon transmission of the laser pulse, wherein a clock is used to drive the described acquisition. The arrival times of photons are divided into time bins corresponding to different ranges. Each "tick" of a digital counter/clock defines a time bin having a width defined by the clock frequency. The comparator based threshold detection operations and corresponding usage of shift registers to capture the comparator outputs in this document are for a voxel intensity determination. The shift registers in this apparatus are operated by a clock.
U.S. Patent No. 6,108,071 illustrates the use of high-speed clock/counter circuitry for determining laser pulse times-of-flight. In the '071 patent, the number of clock pulses that lapse between the emission of a laser pulse and the detection of the corresponding return pulse yield a "coarse" elapsed time measurement. The '071 patent uses a small digital delay line, e.g., ten taps, to detect fractional clock intervals between the actual time of laser pulse emission and the next clock cycle, and between the actual time of return pulse detection and the next clock cycle. These fractional times are added to the coarse time, to yield a more accurate timeof-flight measurement.
What is needed is a high speed time interval measurement system for use in applications such as time-of-flight systems, especially in systems used to guard machinery or the like. Such measurement system should be inexpensive to fabricate, preferably using existing CMOS processes, should exhibit low power consumption, and should provide timing and strength information for one or more return pulses. Such measurement system should rapidly detect multiple return pulses, preferably within time intervals of less than about 500 ns, with a sub-nanosecond timing resolution that can provide spatial resolution of ±5 cm or less. Further, the system should measure return pulse signal strength with sufficient precision for use as an index to a lookup table to correct for timing walk. The system should communicate range measurements with a minimal amount of data. Finally, the system should exhibit reduced sensitivity to variations in ambient temperature, operating voltage, and fabrication processes.
The present invention provides such a high speed measurement system.

SUMMARY OF THE INVENTION

The present invention provides a high speed time interval system to measure time intervals Δt in time-of-flight measurement systems, preferably for use in systems that guard the perimeter of machinery or the like. The system measures time interval between a transmitted scanned laser pulse and a return pulse to determine distance, and can be fabricated on an integrated circuit (IC) using generic components. But unlike prior art systems, the system also returns a measurement of the width of the detected return pulse. Such information is used as indices to a lookup table that stores time walk corrections to the measured range distance. Knowledge of the return pulse width permits inferring strength of the return pulse, which inferred strength is used to estimate time walk Tw. Time walk Tw represents systematic error arising in raw range measurement due to fluctuations in detected signal strength and timing uniformities, the latter arising from process-dependent effects. The IC comprising the present invention is realizable with relatively inexpensive CMOS fabrication processes such that multiple data inputs may be incorporated into the IC without incurring significant additional cost.
The present invention is used with a zone protection system that includes a laser transmitter and photo detector that together define a coaxial field of view. A motor and mirror assembly cause emitted laser pulses and the detector field of view to scan a two-dimensional protection zone. The laser is pulsed with a specified frequency and in synchronously with motor-mirror rotation. A reference target is disposed within the zone protection system housing to reflect a portion of the transmitted energy back to the photo detector for use in compensating the present invention against system thermal drift.
In making time measurements, the present invention eliminates high speed clocks and high speed digital counters such as are commonly employed in prior art time-of-flight measurements. Instead, the present invention derives a START pulse from the laser drive signal (LASER START). This pulse is input to a latch whose output signal is propagated through a tapped delay line portion of a tapped delay line register (TDLR). The tapped delay line comprises preferably 512 buffers that each contribute an incremental time delay to the latch signal.
The detected return pulse is coupled to a bank of comparators (e.g., four comparators denoted CH0, CH1, CH2, CH3) that test the pulse against different threshold magnitudes. The comparator bank output signals are input to the TDLR, namely to individual channel registers that are also coupled to the preferably 512 delays from the tapped delay line register. The TDLR also receives the detector signal returned from the reference target. The TDLR is coupled to a microprocessor for readout, the microprocessor having access to a look-up table (LUT) that contains timing walk information that can be used to correct system errors.
The latched START pulse functions as a clock signal for the TDLR and clocks the detection data output from the comparator bank. The START-clocked TDLR, which preferably has four data inputs, functions in a manner similar to a two-bit transient recorder that clocks synchronously with the LASER START signal and can provide four signal values. For each data channel, the TDLR outputs a first over threshold (FOT) signal proportional to the rising edge of the first detected return pulse. FOT may be treated as the first O-to-1 transition of the detected pulse signal. The FOT enables the microprocessor to provide a measure of time delay and thus of distance X to the target that returned the detected pulse. By itself the FOT can thus serve to approximate distance X. However the TDLR further determines and outputs a total time over threshold signal (TOT) that is proportional to how long the detected pulse energy exceeded a threshold. The TOT may be considered as how long the detected signal, after transitioning O-to-1 remains at 1 before transitioning from 1-to-0. The TOT information permits the microprocessor to infer strength of the returned pulse, including rising and fall transition slopes and pulse width. The inferred strength information is used to index previously stored timing correction information in the look-up table. Such timing correction information will have been obtained during the system design by examining timing walk behavior of the system. The look-up table permits correction, as needed, to the FOT information. Further, the elimination of high speed clocks and high speed counters permits the system to be battery operated and fabricated as a single integrated circuit.
Other features and advantages of the invention will appear from the following description in which the preferred embodiments have been set forth in detail, in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 depicts a generic time-of-flight measurement system, according to the prior art;
FIG. 2 depicts a time-of-flight interval measurement system, according to the present invention;
FIG. 3A depicts a multi-channel comparator block for use with a time-of-flight interval measurement system, according to the present invention;
FIG. 3B depicts exemplary comparator channel signals and START signal as a function of time, according to the present invention
FIG. 4 is an overview block diagram of TDLR 230, according to the present invention;
FIG. 5 is a block diagram depicting details as to the channel registers and test register shown in Fig. 4, according to the present invention;
FIG. 6 depicts timing requirement for a preferred embodiment of a tapped delay line register, according to the present invention; and
FIG. 7 depicts START pulse matching, according to the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Fig. 2 depicts a time-of-flight measurement system 100, according to the present invention. Some components of system 100 may be similar to the prior art system 10 described with respect to Fig. 1. In response to receipt of a current trigger pulse, pulse generator 110 outputs pulses (LASER START) to a laser or other light emitter 120. In response, transmitter 120 transmits energy pulses of perhaps 3 ns to 4 ns, measured at full pulse width, half magnitude. The LASER START pulse is passed through a comparator to form a START pulse that typically commences some tens of nanoseconds (ns) after commencement of the LASER START pulse.
A lens or other mechanism 130 may be used to collimate or focus the transmitter output energy, shown as idealized pulses 140 having amplitude P1, upon a target located at some distance X from lens 130. In a preferred zone protection system, distance X will represent the perimeter distance within which no intrusion is acceptable.
Some emitter optical energy will strike the target (TARGET), e.g., an intruder within the perimeter range, and will be returned toward system 100. The return energy, shown as idealized return pulse train 150 having amplitude P1' enter a lens or the like 160 and are sensed by a detector 170, which preferably comprises avalanche or photo diode detection elements.
Transmitter 120 and detector 170 are configured to provide a coaxial field of view. A motor and mirror assembly 180 functions to scan the coaxial field of view in a plane through an arc defining the zone to be protected. A reference target 190 is disposed within the housing containing at least the transmitter, detector, motor and mirror components such that a reference or test signal may be returned from target 190 for use in correcting distance measurements made by the overall system. Collectively, the above-described units may be referred to a system 200, which system generates a START signal and detection signal (DATA) used by the present invention, which is shown as system 210.
According to the present invention, a time-of-flight time interval measuring system 210 includes a bank of comparators 220, a tapped delay line register 230, a microprocessor 240, and at least one look-up table 250-1, 250-2, etc: Comparator bank 220, shown in detail in Fig. 3A, preferably includes several channels of comparators that compare the incoming DATA signal against separate signal threshold levels. In the preferred embodiment, comparator bank 220 comprises four comparators that output four channels of signal, denoted CH0, CH1, CH2, CH3. Other systems could, however, use fewer or greater than four channels of comparison. Tapped delay line register (TDLR) receives as input the START signal, as well as the preferably four channels of comparator signals, CH0, CH1, CH2, CH3. A detector signal (TEST) derived from the reference target 190 may also be coupled as input to the TDLR.
As described in detail with respect to Fig. 4, TDLR 230 propagates the START signal through a latched tapped delay line, and passes the preferably four channels of DATA through four separate channel registers. The output of the channel registers are multiplexed and decoded. The TDLR outputs DONE, FOT, and TOT signals to a microprocessor 240. A look-up table 250-1 is coupled to the microprocessor. The raw FOT signal alone will enable the microprocessor to calculate distance X to the target returning the detected pulse. However, advantageously the TOT signal characterizes the return pulse and permits the microprocessor to correlate information in the look-up table to make corrections, as needed, for system error in the distance X calculated form the raw FOT signal. If desired, all of system 210 may be fabricated on a single integrated circuit chip, and thus reference numeral 210 may also be understood to refer to the integrated circuit chip. Power consumption for system 210 is sufficiently low, given that no high speed clocks and counters are used, that system 210 may be operated from a battery power source B1, for example about 5 VDC.
Fig. 3A depicts an exemplary comparator block 220, in which four channels of comparator output are provided: CH0, CH1, CH2, CH3. An analog signal (DATA) is coupled as input to the block of comparators, each of which preferably has its own separate threshold voltage level, e.g., Vth0, Vth1, Vth2, Vth3. Fig. 3B is a timing chart depicting the output channels from block 220 and the START signal as a function of time. All time measurements are relative to the START signal, and indeed the START signal commands the TDLR to begin acquiring DATA.
The use of multiple comparators as contrasted with a single comparator advantageously permits a richer definition or characterization of the total shape of the analog-like detected data pulse. For example, in a single channel system comprising only CH0, FOT might be defined as occurring at 1300 ps from beginning of START, e.g., when the detected signal first transitions 0-to-1. (Of course in a complementary system, transitions might be reversed.) TOT based upon CH0 information alone might be 3200 ps, e.g., 4500 ps - 1300 ps, or the time CH0 remains at 1 having transitioned 0-to-1. It is seen from Fig. 3B that a sampled snapshot-like image of the DATA waveform in question is captured. If the target from which the acquired data were returned were highly reflective to the transmitted energy, the pulse-like shape observable in Fig. 3B would likely be "fatter" in that CH1, CH2 and CH3 data might look somewhat more similar to CH0 data. This ability to learn more about the target from the shape of the return signal is an advantage provided by the present invention.
It will be appreciated that multiple channels of sampled data also permits parameterization with different comparator thresholds. This flexibility is especially useful where system non-linearity is present and improved measurement precision can result from examining data obtained with a given threshold or thresholds. Further, in addition to providing the ability to measure slightly different information, multiple channels of sampled data provides redundancy. For example ageing of system components on the integrated circuit containing system 210 may be detected by inputting the exact same information to CH0, CH1, CH2, CH3 and determining when over the lifetime of the channels a substantial error begins to appear.
Fig. 4 provides a somewhat more detailed overview of TDLR unit 230. TDLR 230 provides a channel register for each channel of data output by comparator bank 220. Thus, after passing through a buffer 260, CH0 data is passed to a preferably 512-input channel register 270-0, CH1 data is passed to 512-input channel register 270-1, and so forth. As described later herein, each channel register synchronously captures and stores sampled channel signals preferably in 512 D-type flipflops.
The START signal (typically ≥ 2 ns wide) represents a start-enabling signal for TDLR 230 and is passed through a latch 300 and into a delay line 290. Latch 300 prevents multiple sampling of the channel data due to any jitter on the START signal. Delay line 290 preferably comprises a series-coupled chain of buffers, e.g., 512 buffers, each of which contributes a single buffer nominal delay of about 200 ps. The individual buffers ideally are identical but in practice are substantially identical, which is to say their individual delays may have a tolerance of up to about ±10%, rather than an ideal ±0%. The output from each of the 512 buffers is coupled to test register 280 and to each of the channel registers 270-0 through 270-3. As described later herein with respect to Fig. 6, preferably a crystal controlled phase lock loop (PLL) controlled VCO circuit 315 functioning at about 49 MHz is included to help compensate for thermal drift within delay line 290 by injecting, as needed, compensation current I_COMP. This clock is the only master-type clock found in system 210.
The preferably 512 output lines from each channel register 270-0 through 270-3 are coupled as input to a multiplexer 320. Outputs from test register 280 are coupled as input to a decoder 310, which decoder is enabled by a signal from latch 250. Multiplexer 320 and decoder 310, and delay line 290 output the signals indicated, which signals are coupled to microprocessor 240.
Functionally, the channel registers serve to synchronously sample the incoming CH0, CH1, CH2, CH3 signals with a sample window of approximately 100 ns. Sampling commences with the arrival of the START signal, a O-to-1 transition being used in the preferred embodiment. In the preferred embodiment, latch 300 cannot be cleared while START remains high at logical 1. A DONE signal is generated by TDLR 230 when the sampling period has terminated and all samples have been taken. This operation is quite different than prior art approaches in which a start pulse would enable a high speed counter or would toggle a high speed clock, and in which a pulse similar to DATA in the present invention would be used as a stop pulse to disable the counter or high speed clock. It will be appreciated that in the present invention, there is no master clock or pulse generator to be stopped.
Microprocessor 240 reads data from TDLR 230, preferably via a 16-bit bus, wherein each data address corresponds to sixteen individual data samples. Each of the four comparator channels preferably has a base address, and offsets to the base address correspond to sequential periods of time for that channel, as captured by the TDLR, which functions as a sampling block. Within a data word, bit 0 corresponds to an earliest sample, while bit 15 corresponds to the latest sample.

Microprocessor 240 preferably permits serial writing of test vectors to test register 280. When a write is made to a specified address, the least significant bit (LSB) of the DATA bus is read into the first position of the test register, which corresponds to the earliest time slice. The remainder of the test register is shifted, and the bit corresponding to the latest time slice is discarded. During write cycles, the following addresses command TDLR 230 to function as noted, although other mode commanding protocols could instead be adopted:

XX00000	load data from DATA bus into configuration register
XX00001	reset START latch to enable TDLR to acquire new data
XX00010	shift LSB of DATA to test register
XX00011	clear test register

Reference is now made to Fig. 5 in which further details of channel registers 270-0 to 270-3, and tapped delay line 290 are shown. TDLR sampling is based upon timing signals available from tapped delay line 290, which preferable comprises a series-connected chain of 512 digital buffers. Rather than use an explicit clock signal, TDLR 230 uses sampling signals available from the individual buffers, each of which contributes a time delay. As noted above, the START signal sets a preferably set-reset (SR) latch 300, and the latch output is coupled to the first input of the tapped delay line 290, and will thus self-propagate through delay line 290.
The matrix comprising tapped delay line 290 preferably is 512 x 4 channels. Typically the FOT and LUT values are subject to microprocessor 240 correction since timing imperfections can exist within the present invention due, for example, to non-uniformity of each buffer delay stage and buffer location. The present invention tries to compensate for buffer-to-buffer variations that are a function of physical location within the array. Eight columns of buffer taps are used, comprising buffers 1-64, 65-128, 129-192, 193-256, and so forth. The various taps can each provide about 200 ps to 220 ps variation. After fabrication of integrated circuit 210, a characterization of timing uniformity may be performed, and at least one look-up table (LUT) 250-1, 250-2, etc. is provided containing time corrections for the various buffers; see Fig. 2. Note that the buffer time correction LUT is not the same LUT that is used for the for the timing walk.
As shown in Fig. 5, preferably a delay component (DELAY ALIGN) is inserted in series between the START pulse and the SR latch to force matching of time delay between the four channel inputs CH0 to CH3 and the START input. As noted, a DONE signal is output from the last buffer (e.g., buffer 511 in the preferred embodiment) comprising tapped delay line 290.
An individual one of the preferably 512 buffers comprising tapped delay line 290 may be referred to as CB(T), where T is a number corresponding to the buffer position within the chain of buffers, 0≤ T ≤511. Using this nomenclature, START is input to CB(0), and DONE is output from CB(511).
As shown by Figs. 4 and 5, test register 280 is disposed between tapped delay line 290 and the four channel capture registers 270-0 to 270-3. Test register 280 provides a mask to control sampling of the capture registers, and for every CB(T) tap output from delay line 290 there is one flipflop in test register 280. As shown in Fig. 5, preferably the tap from the delay line 290 and the output from a flipflop in test register 280 are input to a NOR gate whose output is coupled to the clock inputs of four corresponding D-type flipflops, one flipflop in each of the channel registers 270-0 through 270-3.
If the flipflop in the test register holds a 0, then the NOR gate output transitions O-to-1 when the tapped signal transitions 1-to-0. If the test register flipflop holds a 1, the NOR gate output is held 0. Test register 280 preferably is a shift register with new data loaded into the bit that controls samplings of the earliest time slices of the four CH0, CH1, CH2, CH3 signals. Data loaded to the test register may be controlled from the DATA bus, or from external pins.
As described early and as shown in Fig. 5, D-type flipflops in the four channel capture registers 270-0 through 270-3 capture every sample acquired by TDLR 230 synchronously with the START pulse. An individual D-type flipflop (DFF) may be referred to as DFF(N,T), where N and T are integers 0≤N≤3, and 0≤T≤511 that respectively represent the channel register to which the DFF belongs and the section of the overall sampling period captured by that flipflop. Thus, the output of DFF(2,55) denotes the state of CH2 when the 55th sample occurred. The clock (CLK) inputs of DFF(0,T), DFF (1,T), DFF(2,T) and DFF(3,T) are coupled to the output of CB(T), and the D input of DFF(N,T) is coupled to the ChN input to the TDLR system.
In the preferred embodiment, every data sample captured by TDLR 230 preferably is addressable as part of a 16-bit data word, and as shown in Fig. 4, multiplexer, decoder, and associated combinatorial logic components are used to place values from the various D-type flipflops on the DATA bus. The protocol used was described earlier, namely each address corresponds to 16 individual samples identifiable with channel base and time period offset values. Thus, address 0x55 results in the DATA(0:15) outputs corresponding to D-type flipflops (2,366:351).
Preferably during a write cycle with address 00000, values from the DATA(2:1) bus are written to a configuration register, whose three-bits are used as follows:

Config(1): F_OUT_DIS disables output frequency from the phase lock loop (PLL) associated with VCO 315 in Fig. 4;
0 => the PLL frequency is output to F_OUT pin, whereas 1=> F_OUT pin is high impedance;
Config(2): TEST_SELECT selects whether data and clock for test register 280 shall come from the bus interface or from circuit pins;
0 => test register 280 controlled from bus, whereas 1=> test register controlled from pins.

Preferably TDLR 230 has a global output enable ability such that when this signal is 0, all outputs from the integrated circuit containing TDLR 230 (and associated circuitry) behave as described. However when the global enable signal is 1, preferably DATA and DONE outputs go to high impedance, and the VCO PLL 315 (see Fig. 4) powers down, which causes the F_OUT pin not to be driven.
Fig. 6 depicts a portion of channel registers 270-0 through 270-3, and a portion of delay line tapped register 290, to illustrate timing requirements. Let Ct_prop9=(T) denote the time difference, including delays from propagation and wiring, between the rising edges of the output of adjacent buffer elements CB(T) and CB(T+1) within delay line tapped register 290. Nominally each element DB(T) has a propagation delay of about 200 ps, but the amount of delay per element can vary due, for example, to fabrication differences from element to element, and especially due to thermal drift in the circuitry itself.
In an ideal delay line 290, the accumulated delay after a number N of series-connected buffers would be (N)(200), where 200 ps is the nominal per buffer delay. But fabrication variations of the buffers can result in too much or too little delay for different ones of the 512 buffers, and different physical location on the integrated circuit die on which system 210 is fabricated can also result in delay variations. As noted, before packing the integrated circuit, a calibration pulse is propagated through tapped delay line 290, and the actual delay at various tap points within the series of delay buffers is measured. A correlation between buffer tap location and actual delay can be made, and corrected values, where needed, can be stored, for example in look-up table 250-1, 250-2, etc. for use by microprocessor 240. (see Figs. 2 and 4).
PLL-controlled VCO 315 can control at least some of the supply current (I_COMP) to the buffers comprising delay line 290. PLL 315 can control delay within delay line 290 such that thermal drift effects upon the nominal 200 ps per buffer delay do not vary by more than about ±10%, which delays typically increase with increasing temperature. This level of compensation can be maintained for all process variations including mismatch between PLL 315 and the buffers comprising the tapped delay line register 290. Further, total propagation time from the START input to the DONE output signals in the tapped delay line can be held to within about ±5% from integrated circuit chip to integrated circuit chip.
Referring to Figs. 6 and 7, preferably the individual D-type flipflops, e.g., 330-0, 330-1, etc. comprising the four channels of capture registers 270-0 through 270-3 are laid out on the integrated circuit chip such that a maximum time difference between arrival of a O-to-1 signal edge arriving from a channel input to any of the D-flipflop inputs within a channel is about 10 ps. Likewise, two 0-to-1 edges presented to channel register inputs should arrive at their respective D-type flipflop inputs within a column separated by more than 20 ps. Further, the O-to-1 edge from the NOR gates in test register 280 must reach the CLK inputs of the D-type flipflops with less than about 2 ps time separation.
As shown in Fig. 5, a delay-align element is preferably present between the START pulse and the latch 300 to ensure that two O-to-1 signal edges presented simultaneously at START and at the ChN inputs will fall within about 50 ps of each other upon arrival at the CLK and the D-inputs to DFF(N,0). Within a channel register, all D-type flipflops receive their D input signals from a common source, namely the ChN input. Distribution of this signal preferably holds skew to within 10 ps between arrival of a signal at any D-type flipflop in the same channel register. It will be appreciated that during signal transitions occurring during a time window between setup and hold will not yield reproducible results, and accordingly the setup-hold window differential is less than about 50 ps.
Within the channel registers, four D-type flipflops will sample simultaneously and their outputs are multiplexed to a single signal based upon the two most significant bits of the address. Thus, time from valid address inputs to time of valid output from multiplexer 320 (Fig. 4) is preferably less than about 2.25 ns.
Each signal sample stored in TDLR 230 represents the state of one of the inputs at a given point in time, and it is these states that are readout by microprocessor 240 to determine time of the O-to-1 and the 1-to-0 transitions in deriving FOT and TOT.
Signal measurement error due to granularity results from the fact that signals are measured at finite points, and the separation between measurement points will define the maximum accuracy with which system 210 can detect time of a transition in a signal. Using a nominal 200 ps sampling window interval with a ±10% variation, 220 ps would represent worst case error due to granularity. However microprocessor 240 may assume that the sample corresponds to a point midway between it and the preceding sample such that time-of-flight error due to granularity is about ±110 ps. Setup/hold violation means a transition occurring within the sample window between setup and hold time is essentially meaningless. Using a design goal of a sample window with no more than about 50 ps uncertainty will contribute an additional ±50 ps to granularity. There is also a tolerance associated with characterization, given that the leading edge of a sample occurs within a given channel register bin, whose translation to time cannot be achieved with accuracy greater than the tolerance with which the characterization of the integrated circuit chip is known. The resultant measurement error can readily contribute an additional ±90 ps to time-of-flight measurement error. Finally, uniformity of the distribution of data input to the D-type flipflops is not absolute, and is assured only to within about ±10 ps. Adding the above errors indicates that the present invention can measure time-of-flight within about ± 260 ps.
In the preferred embodiment, system 210 was fabricated on a 4 mm x 4 mm chip (also denoted 210), using a 0.5 µm CMOS process and commercial cell libraries for the various flipflop and buffers. Unlike prior art time-of-flight measurement systems, the present invention may be battery operated in that power consuming and heat generating components such as high speed master clocks and high speed digital counters are not required. A further benefit is that substantially less electromagnetic interference (EMI) is generated by the present invention than by prior art systems, which means a less costly enclosure can suffice, since EMI shielding requirements are less stringent. Further cost to implement the present invention can be substantially less than to implement prior art systems yet is more robust and provides excellent time resolution granularity.
Modifications and variations may be made to the disclosed embodiments without departing from the subject of the invention as defined by the following claims.

Claims

A method of determining a distance to an object, the method including emitting a laser pulse directed towards the object and receiving a return reflection from the object, said method characterized by:
propagating a latched version of a start pulse (START) through a tapped delay line (290) comprising a series-connected chain of digital buffers that each contribute an incremental time delay to the latched start pulse and output a corresponding sampling signal, said start pulse associated with output of the laser pulse;

coupling an analog data signal (DATA) to one or more comparators (220) that each outputs a comparator signal based on comparing the analog data signal to a respective different signal level threshold, said analog data signal including a return pulse representing radiation reflected back from the object;

capturing the state of each comparator signal in a corresponding channel register (270) comprising a set of flipflops that are individually clocked by a respective one of the sampling signals output from the tapped delay line (290), so that the channel register (270) captures the state of the comparator signal at sample times corresponding to the incremental delays of the individual digital buffers in the tapped delay line (290), commencing with the start pulse;

determining a first-over-threshold, FOT, time and a total-over-threshold, TOT, time for the analog data signal from the states of the one or more comparator signals captured by the one or more channel registers (270);

calculating a distance to the object as a function of the FOT time; and

correcting the calculated distance for errors arising from fluctuations in the strength of the return pulse by looking up a timing correction based on the TOT time.
The method of claim 1, wherein looking up the timing correction based on the TOT time comprises inferring a strength of the return pulse from the TOT time and looking up a timing correction from a stored lookup table (250-1) of timing corrections correlated with return pulse strength.
The method of claim 1, further comprising correcting the calculated distance for errors arising from delay non-uniformities of individual digital buffers in the tapped delay line (290) by looking up a timing correction based on the FOT time.
The method of claim 3, wherein looking up the timing correction based on the FOT time comprises looking up a timing correction from a stored lookup table (250-2) of timing corrections correlated with FOT times.
A time-of-flight measurement system (100) for determining a distance to an object, said time-of-flight measurement system (100) including a transmitter (120) configured to emit a laser pulse directed towards the object and a detector (170) configured to receive a return reflection from the object, said time-of-flight measurement system (100) characterized by comprising:
a tapped delay line (290) comprising a series-connected chain of digital buffers adapted to contribute an incremental time delay to a latched version of a start pulse (START) propagating through said tapped delay line and to output a corresponding sampling signal, said start pulse associated with output of the laser pulse;

one or more comparators (220) adapted to output a comparator signal based on comparing an analog data signal (DATA) to a respective different signal level threshold, said analog data signal including a return pulse representing radiation reflected back from the object;

one or more channel register (270) comprising a set of flipflops that are individually clocked by a respective one of the sampling signals output from the digital buffers of the tapped delay line (290), adapted to capture the state of each comparator signal at sample times corresponding to the incremental delays of individual digital buffers in the tapped delay line (290), commencing with the start pulse;

and a microprocessor (240) adapted to determine a first-over-threshold, FOT, time and a total-over-threshold, TOT, time for the analog data signal from the states of the one or more comparator signals captured by the one or more channel registers (270);

to calculate a distance to the object as a function of the FOT time; and to correct the calculated distance for errors arising from fluctuations in the strength of the return pulse, based on looking up a timing correction based on the TOT time.
The time-of-flight measurement system (100) of claim 5, wherein the time-of-flight interval measuring system (210) includes a microprocessor (240) that is configured to calculate the distance to the object based on the FOT time.
The time-of-flight measurement system (100) of claim 6, wherein the microprocessor (240) is configured to look up the timing correction based on the TOT time, based on said microprocessor (240) being configured to infer a strength of the return pulse from the TOT time and look up a timing correction from a stored lookup table (250-1) of timing corrections correlated with return pulse strength.
The time-of-flight measurement system (100) of claim 6, wherein the microprocessor (240) is further configured to correct the calculated distance for errors arising from delay non-uniformities of individual digital buffers in the tapped delay line (290), based on being configured to look up a timing correction based on the FOT time.
The time-of-flight measurement system (100) of claim 8, wherein the microprocessor (240) is configured to look up the timing correction based on the FOT time based on being configured to look up a timing correction from a stored lookup table (250-2) of timing corrections correlated with FOT times.